Qur’an Atlas receiving trial — research record

This record contains the sender answer and rubric. For a new initial interpretation trial, use inputs/receiver-kit.zip and withhold the other material.

Contents
- inputs/receiver-kit.zip: exact original kit sent to the receiver.
- returns/receiver-return.zip: exact returned archive, with code, interpretation, data and first-run history.
- sender/: original USL1, expanded expression, conceptual composition, English reading and rubric from release93.
- publication-audit/: the local technical replay, qualitative semantic assessment and execution logs. This audit is not another external AI interpretation.
- sender-verification/ and published-snapshot/: the earlier publication/review evidence frozen for release93. These prove what the sender checked then, not perpetual website state.
- results.txt and results.json: concise human and machine-readable publication records.
- FILES.json: hashes of every other research-record member.

To reproduce the receiver’s checks
Extract returns/receiver-return.zip into a fresh directory. Enter its usl_quran_atlas_recovery folder and follow METHOD.md. With Python and Node available, the recorded steps are:
python verify_inputs.py
node --require ./offline.cjs receiver/recover.cjs
node --require ./offline.cjs inspect_reference.cjs
python audit_quran.py
python test_audit.py

Those commands regenerate data and reports, not the authored English judgment. Keep the original ZIP unchanged. The Node preload guard is not an OS network sandbox. The original receiver environment’s unavailable unshare command is historical evidence, not a prerequisite to rerun the artifact checks.

The scripts under publication-tools/ and publication-audit/semantic_compare.py are the local USS-workspace audit/build record. They retain their original release93/release94 path assumptions; they are not advertised as standalone installers. The complete returned receiver code above is the portable reproduction route. The original source and lexical dependencies are already inside the receiver and return ZIPs.

One contextual receiver result is recorded. The model label is self-reported. Technical equality, interpretation and agreement are separate findings.
